Eyal,

Are there any plans / ideas from Veritas on improving backup/restore times of filers ?

The only time NetBackup has a performance impact on backup is when there are a very large number of files in the backup (on the order of millions). Currently we get all the NDMP history in DIR & NODE messages which we then have to convert into full paths for our catalog. This can be CPU, RAM, and disk intensive.

We are investigating options for how we might address this issue.

As for restore time there isn't much we can do besides add support for 'direct access restore'. Sounds easy but it isn't. Even when NetApp supports it, it may not be what you want - it will have limitations. It is a hard thing to do with a dump-style backup. Because of the limitations and the NetBackup model, we may not be able to support it reliably. We always take the conservative approach of doing everything we can to make sure you get the correct data and any new optimizations have to be bullet-proof.

In the end, all we can do is support what NDMP and each NDMP server implementation provides to us.

It seems that the tape/backup s/w is not keeping up well with the filers' huge sizes ?

I don't think it is just filers - the whole industry is seeing an explosion of data. Tape drives certainly haven't kept up with disk.

Of course NetBackup has other features that are not available in the NDMP world simply because 1) we have raw access to the disk, and 2) we control the tape format.

What is the best configuration that can give us best performance ?

From a NetBackup perspective the only thing you can do is try to keep the number of files per backup from getting very large. Trouble is that breaking a volume up into different backups has an impact on filer efficiency (unless you use quota trees).

__________________________________________________________________________ Steve Kappel steve.kappel@veritas.com VERITAS Software steve.kappel@iname.com (Personal)